The Role of Kupffer Cells as Mediators of Adipose Tissue Lipolysis

Macrophages were first described as phagocytic cells that defend organisms against pathogens (1). More recently, it has become clear that resident macrophages populate all tissues and exhibit organ-specific functions. For instance, macrophages contribute to thermogenesis regulation in adipose tissue (2), synaptic pruning, and learning-dependent synapse formation in the brain (3, 4), gastrointestinal motility in the muscularis (5), and electrical conduction in the heart (6). These noncanonical activities highlight the functional diversity of macrophages and emphasize their ability to execute tissue-specific tasks beyond host defense.

The liver is a metabolic and immunologic organ. Macrophages are an intrinsic part of a healthy working liver. Although the role of Kupffer cells (KCs) in liver pathologies has been extensively studied, the contributions of these cells to normal liver physiology remain unclear. Recently, studies have suggested that KCs are involved in iron recycling (7). However, in contrast to what is known about macrophage function in other organs, less is known of the specific functions of KCs in the steady-state. Studies have shown that macrophages play vital roles in maintaining system metabolic homeostasis (8). The liver is an organism’s metabolic hub, but whether KCs play an important role in systemic metabolism in the steady-state has remained largely unexplored.

IL-1β is a major mediator of inflammation, and it also performs numerous functions related to host defense mechanisms by regulating the immune system and the neuronal and endocrine systems that interface with the immune system. Studies have established a strong correlation between IL-1β and metabolism (9). Metabolic stress, such as nutrient overload, is thought to be a trigger of IL-1β expression. A critical sensor of nutrient overload is the NLRP3 inflammasome, which processes pro–IL-1β into its active form in various metabolic disorders (10). Although such studies have shed light on the pathological role of IL-1β, little is known about the physiological function of IL-1β. In response to fasting and feeding rhythms, profound changes in the metabolism of immune cells take place. Beyond supplying energy, nutrients can act as signaling molecules that promote the activation of immune cells. Some studies have found that elevated concentrations of glucose and other metabolites drive the production of IL-1β by macrophage (11, 12). Macrophage-derived IL-1β may also stimulate insulin secretion and promote glucose disposal after feeding (13).

In contrast to IL-1β, fibroblast growth factor 21 (FGF21) is reversely regulated by food intake. FGF21 has been reported as an important hormone in the body’s adaptation to fasting. FGF21 also acts as a prolipolysis factor, and its levels can be increased by fasting and reduced by feeding (14, 15). In the fasted state, FGF21 increases adipose lipolysis and hepatic ketogenesis. However, the regulation of FGF21 gene expression is not fully understood. Clinical research has found that an acute inflammatory response suppresses serum FGF21 levels in healthy adult volunteers (16).

In the current study, KCs were found to play an important role in the regulation of adipose lipolysis. KC depletion increased FGF21 expression in hepatocytes, promoted adipose lipolysis, and led to hyperlipidemia and a thin phenotype. KC-derived IL-1β inhibited hepatocyte FGF21 expression and regulated lipolysis. The present study indicated that the role of KCs in the suppression of lipolysis depends on bacterial products and that KCs play a crucial role in systemic metabolism via the gut–liver–adipose axis.

C57BL/6J mice were purchased from Charles River (Beijing, China). CCR2 knockout (KO) mice were purchased from The Jackson Laboratory. CD11b-DTR mice were generous gifts from Dr. W. Honglin (Shanghai Jiao Tong University). FGF21 KO mice were generous gifts from Dr. S. Kliewer (The University of Texas Southwestern Medical Center). The mice were housed in the specific pathogen–free Beijing Institute of Lifeomics animal facility and were exposed to a reversed 12:12 h dark-light cycle with free access to food and water unless otherwise stated. For all experiments, 6–8-wk-old male mice were used. All mouse experiments were approved by the Institutional Animal Care and Use Committee at the Beijing Institute of Lifeomics.

For KC depletion, mice were injected via the tail vein with 100 μl of clodronate liposome (CLO) (clodronateliposomes.org). After 24 h, serum and tissue samples were collected. In one experiment, mice were injected with 100 μl of CLO at days 1, 3, and 6, and serum and tissue samples were collected at day 7. The diet of one mouse group was changed to a high-fat diet (HFD) during KC depletion. For monocyte depletion, 25 ng/g diphtheria toxin (DT) was given to mice by tail vein injection. After 24 h, serum and tissue samples were collected.

One week before KC depletion, an antibiotic “concoction” consisting of vancomycin (10 mg/ml), neomycin (20 mg/ml), metronidazole (20 mg/ml), and ampicillin (20 mg/ml) (all purchased from Sigma) was administered by gavage every 12 h. The gavage volume of 5 ml/kg body weight was delivered with a stainless steel tube.

Blood was sampled from the vein and centrifuged at 3000 × g for 15 min. Serum was frozen at −20°C for later analysis of total protein, glucose, bile acid, bilirubin, triglyceride (TG), total cholesterol, high-density lipoprotein (HDL-C), and low-density lipoprotein (LDL-C) by an automatic biochemistry analyzer (Hitachi 7020; Hitachi High Technologies). Test kits were purchased from BioSino Biotech (Beijing, China).

3T3-L1 cells were maintained as subconfluent cultures in DMEM supplemented with 10% FBS, and differentiation was induced. To differentiate fibroblasts into adipocytes, the cells were grown for 2 d after confluence. The medium was then changed to DMEM containing 10% FBS, 1 μg/ml insulin, 0.1 μg/ml dexamethasone, and 112 μg/ml isobutylmethylxanthine. After 7 d, the cells were maintained in DMEM supplemented with 10% FBS and 1 μg/ml insulin for three additional days. Finally, the medium was replaced with DMEM containing only 10% FBS for 2 d, after which the cells were used.

For analysis of FGF21-stimulated lipolysis, 3T3-L1 adipocytes were cultured in DMEM without FBS. FGF21 was added to the medium at different concentrations. After 6 h, the cell medium was collected and centrifuged at 500 × g. The resulting supernatant was used for glycerol and free fatty acid (FFA) detection.

Gonadal fat pads isolated from male mice were cut into small pieces, and 20 mg was incubated in 100 μl of lipolysis buffer (KRB buffer plus 3.5% fatty acid–free BSA and 0.1% glucose) in a 96-well plate for 2 h at 37°C with rotation at 450 rpm. The culture medium was collected at the indicated time to measure glycerol and FFA contents.

The mice were anesthetized, and the abdomens were surgically opened by a vertical incision. The livers were perfused via the portal vein with HBSS for 6 min, followed by perfusion with digestion buffer containing 0.05% collagenase type IV for 5 min. Livers were then excised and disrupted, and the cell suspension was passed through a 70-μm cell strainer. Parenchymal cells were separated from nonparenchymal cells by centrifugation at 50 × g for 4 min at 4°C. The remaining supernatant was centrifuged at 500 × g for 5 min, and the crude nonparenchymal cells were suspended with 24% iodixanol and overlaid with 17, 11.6, and 0% iodixanol in 1640 medium. After the cells were centrifuged at 1400 × g for 20 min, KCs were isolated at 17/11.6 and 11.6/0% iodixanol interphases.

For the in vivo study, mice were given 1 μg/kg IL-1β through i.p. injection, and tissue samples were collected 12 h later for further detection.

For the in vitro study, hepatocytes and AML12 cells were plated in six-well plates with 5 ng/ml IL-1β. After 6 h, cells were collected for mRNA expression detection.

Hepatocytes and KCs were cultured for 24 h, and the medium was then collected as conditioned medium. Hepatocytes were cultured in six-well plates, and the medium was replaced for 4 h with the conditioned medium from hepatocytes and KCs. Hepatocytes were collected for mRNA expression detection. In one group, KC conditioned medium (KCM) was pretreated with 1 μg/ml anti–IL-1β (R&D systems).

Single-cell suspensions of blood and liver were suspended in FACS buffer, and nonspecific binding to Fc receptors was blocked by incubation with anti-CD16/32 Ab (eBioscience). Cells were washed and then stained with Abs from eBioscience (F4/80, CD11b, and CD45). Flow cytometry was performed on an LSRFortessa (BD), and the results were analyzed with FlowJo (Tree Star).

The mice were anesthetized and perfused with 20 ml of PBS, and liver tissues were removed and fixed overnight in 4% PFA at 4°C. The tissues were rinsed with PBS and embedded in optimal cutting temperature compound. Tissue sections were blocked with 5% BSA for 1 h and then incubated overnight with primary Abs against F4/80 (1:200; Abcam). The sections were washed three times with PBST and were then incubated with Alexa Fluor 488–conjugated goat anti-rabbit IgG secondary Abs (1:500) in blocking solution for 2 h. Then, the sections were washed three times with PBST and mounted with DAPI for nuclear staining.

Total liver or hepatocyte RNA was extracted by TRIzol reagent. Tissues were homogenized with a tissue homogenizer, and mRNA extraction was performed according to the manufacturer’s instructions. The RNA concentrations were determined by nanodrop. cDNA was generated from 1 μg of total RNA using an RT Reagent Kit (Takara), and cDNA was amplified using a Bio-Rad iQ5 PCR machine.

Adipose tissue was minced and homogenized in lysis buffer. The samples were centrifuged at 12,000 × g for 10 min, and the supernatant was collected. The protein content was determined by bicinchoninic acid assay, and 10 μg of total protein was loaded per lane of a 10% SDS-PAGE gel. The separated proteins were transferred to a polyvinylidene difluoride (PVDF) membrane and then blocked with 5% BSA in TBST for 1 h. The blocked membranes were probed overnight with Abs. After the membranes were washed, they were incubated with anti-rabbit or anti-mouse IgG-HRP secondary Abs and then washed again. The membranes were incubated with ECL reagent before exposure to film.

The pLIVE plasmid vector was purchased from Mirus Bio (Madison, WI), and the mouse FGF21 gene was cloned from cDNA sequences of C57BL/6 mice. The FGF21 gene was inserted into multicloning sites of the pLIVE vector using restriction enzyme digestion, and it was confirmed by DNA sequencing. A PBS solution containing 5 μg of plasmid DNA was administered to the mice by hydrodynamic tail vein injection.

The levels of serum and liver FGF21 were detected by FGF21 ELISA (Boster Bio, Wuhan, China), and the level of IL-1β in the condition medium was detected by IL-1β ELISA (eBioscience).

Recombinant human FGF21 was expressed in Escherichia coli, refolded in vitro, and purified to homogeneity by sequential affinity, ion exchange, and size exclusion chromatography.

The FGF21 promoter (−1497 to +5) was amplified from genomic DNA from C57BL/6J mice and cloned into the pGL3 report vector. AML12 cells were transfected with FGF21 luciferase reporter using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). The luciferase assay was conducted using the dual luciferase substrate system (Promega, Madison, WI), and the result was normalized with the internal control Renilla luciferase.

In the current study, all in vitro experiments were conducted a minimum of three times. Statistical analysis was performed by using Student t test or two-way ANOVA, and all results were expressed as the mean ± SD, *p < 0.05, **p < 0.01, ***<0.001, and n.s., not significant, according to two-sided Student t test unless otherwise indicated.

To investigate the physiological role of KCs, KC depletion was performed by CLO administration. Depletion of KCs was confirmed by dramatically reduced expression of the F4/80 macrophage marker gene (Fig. 1A) as well as by ablation of F4/80⁺CD11b^int KCs (Fig. 1B). Importantly, other tissue-resident macrophages were left intact as shown by flow cytometric analysis of macrophages (Fig. 1C) and immunostaining for the F4/80 macrophage marker (Fig. 1D, Supplemental Fig. 1A). Some serum analytes were checked after KC depletion, which related to liver metabolic functions (17). Interestingly, we found that KC ablation led to hyperlipidemia without impacting serum glucose, protein, and bile acid (Fig. 2A). Furthermore, serum bilirubin was decreased in mice treated with CLO (Fig. 2A), indicating that KCs play an important role in clearing senescent RBCs.

As expected, blood monocytes were also ablated by CLO treatment (Supplemental Fig. 2A). To exclude the possibility that the phenomenon described above resulted from the depletion of monocytes, CD11b-DTR mice were used to kill monocytes when given DT, but KCs were retained (Supplemental Fig. 2B). Monocyte depletion did not lead to hyperlipidemia (Supplemental Fig. 2C). CCR2 KO mice were used to exclude the possibility that the phenomenon described above resulted from infiltrating monocytes when KCs were depleted. KC depletion in CCR2 KO mice still led to hyperlipidemia (Supplemental Fig. 2D), indicating that KCs are the main factor in the occurrence of hyperlipidemia.

KCs were depleted for a longer time to exclude transient effect of CLO. One week after KC depletion, mice that received CLO still had hyperlipidemia (Supplemental Fig. 1B). Weight loss was observed in mice treated with CLO and fed regular chow or an HFD (Fig. 2B). The weight of epididymal white adipose tissue (epiWAT) was also reduced in mice treated with CLO when fed an HFD (Fig. 2C). Daily food intake was not affected by KC depletion (Fig. 2D). Taken together, these findings suggest that depletion of KCs leads to hyperlipidemia and weight loss.

Serum TGs mainly exist in very low density lipoproteins of which the secretion and reabsorption require some apolipoproteins for assistance. The mRNA expression of some apolipoproteins was examined, but the expression levels were not changed (Supplemental Fig. 3A). Changes in the expression of Fas, Hmgcr, and Hmgcs2, which mediate fatty acid synthesis, cholesterol synthesis, and ketone body synthesis, respectively, were also measured. Only Hmgcs2 expression was enhanced after KC depletion (Fig. 3A). Serum ketone bodies were higher in KC-depleted mice, as indicated by higher serum β-hydroxybutyrate (BHB) levels (Fig. 3B).

FFAs are the major source of ketone bodies (18). KC ablation led to higher serum glycerol and FFA levels (Fig. 3C). The possible role of blood monocytes was excluded, as monocyte depletion using DT did not influence serum glycerol and FFAs (Supplemental Fig. 3B). The possible role of monocytes recruited to the liver was also excluded, as depletion of KCs in CCR2 KO mice led to higher serum glycerol and FFAs (Supplemental Fig. 3C).

Adipose tissue lipolysis supplies FFAs as fuel to tissues and organs and is the primary source of serum FFAs. To determine the role of KC depletion in adipose lipolysis, adipose tissue was isolated for lipolysis detection. White adipose tissue (WAT) from KC-depleted mice showed increased glycerol and FFA release (Fig. 3D), suggesting that KC depletion promotes lipolysis. Furthermore, the ablation of KCs promoted a high level of cAMP in adipose tissue (Fig. 3E), which activates protein kinase A (PKA) and subsequently activates lipases found in adipose tissue (19). In line with these data, KC-deficient mice had a high level of phosphorylated hormone-sensitive lipase (pHSL) at Ser⁵⁶³ and Ser⁶⁶⁰ but not the inhibitory Ser⁵⁶⁵ (Fig. 3F). The expression levels of HSL and adipose TG lipase (ATGL) were not changed. In summary, KC depletion resulted in enhanced lipolysis in adipose tissues.

The mechanism of KC regulation of adipose lipolysis was next investigated. We developed an ex vivo assay in which adipose tissue was incubated with plasma from KC-depleted mouse models, and glycerol and FFAs were quantified as lipolysis markers. We found that plasma from KC-depleted mice promoted adipose lipolysis (Fig. 4A), and plasma from KC-depleted mice enhanced HSL phosphorylation at Ser⁵⁶³ (Supplemental Fig. 4A).

Next, the potential factors that might impact lipolysis and ketogenesis were investigated. The mRNA expression levels of several hepatokines were measured, and FGF21 expression was upregulated in KC-depleted mice (Fig. 4B). In contrast to the FGF21 level in control mice, mice that underwent KC depletion had higher levels of serum or plasma FGF21 (Fig. 4C, Supplemental Fig. 4B) and liver FGF21 (Fig. 4D).

We next tested whether FGF21 can promote lipolysis via in vivo and in vitro lipolysis assays.

To investigate the direct influence of FGF21 on adipocytes, 3T3-L1 adipocytes were stimulated with FGF21; in line with earlier publications, FGF21 promoted 3T3-L1 adipocyte lipolysis but to a lesser extent than isoproterenol (Fig. 4E) and enhanced phosphorylation of HSL at Ser⁵⁶³ and Ser⁶⁶⁰ (Fig. 4F) (20, 21). FGF21 enhanced ERK phosphorylation in 3T3-L1 adipocytes (Fig. 4H), but the prolipolysis effect was not dependent on ERK signaling, as suppressing ERK phosphorylation using MAP kinase (MAPKK) inhibitor (PD98059) did not attenuate the prolipolysis effect of FGF21 (Fig. 4G, 4H).

To investigate the in vivo function of FGF21, the FGF21 expression vector was administered to mice through i.v. hydrodynamics-based injection to induce hepatic FGF21 expression (Supplemental Fig. 4C). Mice with FGF21 overexpression showed a lean phenotype (Supplemental Fig. 4D) and had decreased body weight (Fig. 5A) and adipose weight (Fig. 5B) compared with control mice. However, FGF21 overexpression increased food intake (Supplemental Fig. 4E), suggesting that the weight loss effect was caused by enhanced metabolism. FGF21-treated mice had markedly smaller white adipocytes than control mice (Fig. 5C). Adipose tissues were obtained for lipolysis detection, and FGF21 overexpression markedly increased glycerol and FFA secretion from adipose tissue (Fig. 5D). The lipolytic response was accompanied by increased phosphorylation of HSL at Ser⁵⁶³ and Ser⁶⁶⁰ (Fig. 5E). Moreover, KC depletion did not result in hyperlipidemia (Fig. 6A), promote lipolysis (Fig. 6B), or promote HSL phosphorylation (Fig. 6C) when FGF21 was overexpressed, because FGF21 overexpression eliminated the effect of KCs on hepatocyte FGF21.

FGF21 KO mice were used to further research the role of FGF21 in the KC depletion-related phenotype. FGF21 KO mice and wild-type mice had similar food intake (Supplemental Fig. 4F), which agreed with a previous study (22). KC depletion did not result in hyperlipidemia (Fig. 6D) or promote lipolysis (Fig. 6E) in FGF21 KO mice. Thus, KC depletion led to enhanced expression of FGF21, which may promote adipose tissue lipolysis.

Serum FGF21 has been found to be suppressed by inflammatory stimuli (16). An inflammatory cytokine, IL-1β, was reduced by KC depletion in liver (Fig. 7A). We next investigated whether IL-1β inhibits hepatic FGF21 expression. In an in vitro study, hepatocytes or AML12 cells were stimulated with IL-1β, resulting in IL-1β–induced suppression of FGF21 mRNA expression (Fig. 7B, 7C). A luciferase assay showed that IL-1β inhibited FGF21 expression (Fig. 7D). In an in vivo study, IL-1β was administered to mice through i.p. injection. As expected, IL-1β administration elevated the serum IL-1β levels (Fig. 7E), suppressed hepatic FGF21 expression (Fig. 7F), and reduced the serum FGF21 levels (Fig. 7G). Accordingly, IL-1β administration attenuated adipose lipolysis (Fig. 7H) and reduced the phosphorylation of HSL at Ser⁵⁶³ and Ser⁶⁶⁰ (Fig. 7J).

To directly investigate the suppressive role of KCs on hepatocyte FGF21 expression, primary hepatocytes were stimulated with conditioned media from hepatocytes and KCs (KCM). KCM suppressed FGF21 expression in primary hepatocytes (Fig. 7I), whereas anti–IL-1β Ab preincubation reversed the suppressive effect of KCM (Fig. 7I). In summary, KC-derived IL-1β may suppress FGF21 expression in hepatocytes.

We next investigated whether LPS has the same effects as IL-1β. Acute i.p. injection of LPS promoted IL-1β expression (Fig. 8A), suppressed FGF21 expression (Fig. 8B), and reduced serum FGF21 (Fig. 8C).

Finally, we assessed the role of bacterial products in the stimulation of IL-1β by treating mice with broad-spectrum antibiotics (Abx). Mice treated with Abx had low body weights (Fig. 8D). KCs from Abx-treated mice exhibited low IL-1β secretion (Fig. 8E). However, hepatocytes from Abx-treated mice exhibited higher FGF21 expression (Fig. 8F). Mice treated with Abx had elevated lipolysis, and KC depletion in Abx-treated mice did not enhance lipolysis (Fig. 8G). Food intake affects gut microbiota at physiological condition. The liver mRNA expression of IL-1β and FGF21 was measured during feeding-fasting cycle. We found that feeding promoted IL-1β expression but suppressed FGF21 expression in the liver (Fig. 8H). In summary, microbiota may play an important part in how KCs suppress adipose lipolysis.

The present study showed that KCs in the liver play an important role in the regulation of adipose lipolysis. Gut microbiota may prime KCs to produce IL-1β. KC-derived IL-1β can suppress FGF21 expression in hepatocytes. FGF21, an endocrine hormone, will affect adipose tissue to promote lipolysis. The gut–liver–adipose axis coordinately maintained metabolic homeostasis (Fig. 8I).

The liver harbors the largest proportion of macrophages among all solid organs in the body (23). Thus, liver macrophages have crucial functions in maintaining homeostasis for the liver itself and the whole body. KCs are seeded along sinusoidal endothelial cells and are important scavengers that are constantly clearing gut-derived pathogens from the blood. Studies have shown that KCs are involved in maintaining functional iron metabolism and bilirubin metabolism during homeostasis (7, 24). CLO has been widely used to study the role of KCs in metabolic disease, and CLO can successfully deplete KCs without affecting other liver cells, such as hepatocytes, endothelial cells, and hematopoietic stem cells (25). In the current study, tail vein injection of CLO depleted KCs, which induced weight loss (Fig. 2B). Depletion of KCs led to increased serum TG, FFA, and glycerol levels (Fig. 2A). β-Hydroxybutyrate, a product derived from FFAs, was also increased by KC depletion (Fig. 3B).

Plasma FFA flux is related to whole-body lipolysis. Lipolysis mainly occurs in adipose tissue. During fasting, exercise and aggression, increased lipolysis, and FFA flux bring energy to non–glucose-dependent tissues. Adipose lipolysis is controlled not only by adipose signaling but also by liver signaling. Liver adipose tissue cross-talk contributes to lipolysis and energy supply (26). In the current study, depletion of KCs resulted in enhanced lipolysis in adipose tissue (Fig. 3D), supporting the concept of liver and adipose cross-talk.

Although the liver is commonly overlooked in contemporary obesity research, it is emerging as a central regulator of whole energy homeostasis. Liver-derived proteins, known as hepatokines, are now considered attractive targets for the development of novel obesity treatments (27). FGF21 represents the major hormone secreted from the liver. Some studies have established FGF21 as an important hormone in the intermediate time period in the body’s adaptation to fasting. FGF21 increases adipose tissue lipolysis, hepatic ketogenesis, and circulating β-hydroxybutyrate but reduces glucose levels (14, 15). Several lines of evidence have indicated that FGF21 induces ketogenesis by stimulating lipolysis, thereby increasing the supply of FFAs to the liver. First, FGF21-overexpressing mice had markedly smaller white adipocytes than wild-type mice (Fig. 5C), suggesting that the lipids in adipocytes were depleted. Second, FGF21-overexpressing mice had significantly increased adipose lipolysis (Fig. 5D). Third, FGF21 enhanced lipolysis in 3T3-L1 adipocytes (Fig. 4E), suggesting that FGF21 acts directly on WAT to stimulate lipolysis.

There are some conflicting studies related to the role of FGF21 in regulating lipolysis. Two studies have reported the direct influence of FGF21 on adipocytes. One study has revealed that lipolysis in adipocytes is attenuated by FGF21 treatment for 3 d (28). Another study found that lipolysis in adipocytes is promoted after treatment with FGF21 for 6 h (21). Therefore, the incubation time may impact the results. In vivo studies mainly support the idea that FGF21 promotes lipolysis. Li et al. (29) speculated that FGF21 inhibits lipolysis of adipocytes, but Li’s conclusion was based on serum FFA levels without detecting lipolysis in adipocytes. Hotta et al. (20) indicated that FGF21 stimulates lipolysis in WAT during feeding but inhibits lipolysis during fasting. This conclusion was based on HSL mRNA expression levels and not protein levels. Another study reported that FGF21 attenuates hormone-stimulated lipolysis (30), but these results cannot be used to elucidate the physiological role of FGF21. Using FGF21-overexpressing mice, Inagaki et al. (15) found that FGF21 promotes lipolysis. Using mice in which the βKlotho gene is disrupted in the hypothalamus, Owen et al. (31) found that FGF21 acts on the hypothalamus to induce corticortropin-releasing factor and to stimulate sympathetic nerve activity, which induces uncoupling protein 1 and lipolysis in brown adipose tissue. PF-05231023, a long-acting FGF21 analogue, decreases body weight in obese nonhuman primates and humans (32). These data all support the idea that FGF21 promotes lipolysis. The results of studies using tissue-specific ablation of FGFR1 or KLB in obese animal models support the role of FGF21 in promoting lipolysis (15, 33, 34).

Stienstra et al. (35) found that KC depletion mainly reduces the expression of macrophage-related genes, including IL-1β, according to microarray analysis. In the current study, the gene expression of some hepatokines was measured, and FGF21 expression was increased. We believe that more hepatic pathways must be affected after KC depletion. DNA microarray or proteomic strategies may be needed to elucidate the extensive change in the liver or hepatocytes after KC depletion. Recently, some new hepatokines that mediate metabolism in adipose tissue have been reported. Tsukushi is a liver-enriched secreted factor, and its deficiency may lead to increased sympathetic innervation and norepinephrine release in adipose tissue, leading to enhanced adrenergic signaling and lipolysis (36). Activin E functions as a hepatokine that stimulates energy expenditure through brown and beige adipocyte activation (37).

ATGL and HSL are the major rate-determining enzymes of lipolysis. The expression profile of HSL basically mirrors that of ATGL, given that both enzymes coordinatedly hydrolyze triacylglycerol (TAG) and, therefore, share some regulatory characteristics but differ in the mechanisms of enzyme control. Although β-adrenergic stimulation regulates ATGL primarily via recruitment of the CGI-58 coactivator, HSL is a major target for PKA-catalyzed phosphorylation (38). In the current study, no change in expression of total ATGL and HSL was detected. Perhaps HSL is more sensitive to FGF21 signaling, suggesting that the phosphorylation and activity of HSL may be enhanced by KC depletion.

In the current study, KC depletion promoted FGF21 expression (Fig. 4B–D). Prior research has shown that IL-1β derived from KCs suppresses proliferator-activated receptor α (PPARα) expression and activity, whereas KC depletion increases the expression of genes involved in fatty acid oxidation, including PPARα (35). Hepatic gene expression of FGF21 is regulated by PPARα in response to starvation (14). We speculated that IL-1β from KCs suppresses FGF21 expression. The present study showed that IL-1β inhibited hepatocyte FGF21 expression both in vivo and ex vivo (Fig. 7B, 7C, 7F, 7G). Consistent with this finding, IL-1β also inhibited adipose lipolysis (Fig. 7H). However, IL-1β was given i.p. in this experiment and can affect adipose lipolysis in many pathways. For example, IL-1β can directly impact on the adipocyte, to promote lipolysis (39). Peritoneal macrophage-derived IL-1β can act on the pancreatic B cells to stimulate the secretion of insulin (13). These effects of IL-1β could not be ruled out. In the present experiment, IL-1β was given for 12 h. Meantime, the mice were fasted overnight to promote FGF21 expression and lipolysis. Fasting status may lead to the antilipolytic effect of IL-1β in our experiment and may be the cause of inconsistency, whether IL-1β stimulates or inhibits adipose lipolysis. Mice with KCs that specifically lack IL-1β should be used in future study to elucidate the role of KC-derived IL-1β in adipose lipolysis.

Because of its anatomical position and unique vascular system, the liver is susceptible to exposure to the microbial products from the gut. Microbial products in the portal venous blood eventually reach the sinusoids in the liver, where KCs reside. Microbial products, including LPS, induce KCs to secrete IL-1β (40, 41). In the current study, bacterial clearance reduced IL-1β secretion of KCs and led to enhanced lipolysis and weight loss (Fig. 8D, 8E, 8G). We found that FGF21 as a prolipolytic hormone was increased after the drop of IL-1β (Fig. 8F). Elevated FGF21 may be one pathway lead to lipolysis in Abx-treated mice. Other pathways also exist. For example, depletion of the microbiome leads to reduced absorption of nutrients that will explain the reduction in body weight gain and leads to adipose tissue lipolysis (42, 43). KC depletion could not result in enhanced lipolysis in Abx-treated mice, indicating that microbiota may play a role in enhanced lipolysis related to KC depletion (Fig. 8G). Direct evidences are needed in future study to testify the role of interaction between microbiome, KCs, and hepatocytes in lipolysis. Crawford et al. (44) reported that germ-free mice have less FGF21, which may be caused by the total loss of bacteria. Siegfried et al. showed that the FGF21 levels positively correlated with Lactobacillus reuteri and negatively correlated with Clostridium lactatifermentans (45).

To adapt to the daily cycles of nutrient availability, energy metabolism in animals has evolved to be cyclical. The feeding-fasting cycle is an important part of the metabolic cycle. After a meal is eaten, dietary TG is hydrolyzed by lipase to yield fatty acids and monoacylglycerol, which are then absorbed and stored as TG in adipose tissue. During fasting, TG in adipose tissue is hydrolyzed to provide energy. Although this process is regulated mainly by insulin, other factors may also participate. The present study measured IL-1β and FGF21 expression in fasted and nonfasted mice. Feeding promoted IL-1β expression but suppressed FGF21 expression in the liver when it was time to store energy (Fig. 8H). When KCs were depleted, IL-1β expression was decreased, FGF21 expression was increased, and adipose lipolysis was promoted even it was not needed to supply energy (Supplemental Fig. 3D).

CLO has previous been used to deplete adipose tissue macrophages by peritoneal injection, demonstrating that adipose tissue macrophage are related to lipolysis (46, 47). In the current study, i.v. injection of CLO did not impact adipose tissue macrophages. Adipose tissue is an important endocrine organ responsible for systemic metabolic regulation. Regulatory peptides released from adipose tissue (adipokines) have been shown to exert defined effects on the liver (26). A previous study has found that the insulin-sensitizing adipokine adiponectin is a downstream effector of FGF21. FGF21 enhances both the expression and secretion of adiponectin in adipocytes, thus mediating the systemic effects of FGF21 on energy metabolism and insulin sensitivity (48). Therefore, adipose tissue and the liver form a metabolic unit that coordinately maintains metabolic homeostasis.

In conclusion, the present findings showed that KC depletion leads to promoted adipose lipolysis, indicating a lipolysis regulatory role of KCs. Our findings also indicate that intraorgan cross-talk (KCs and hepatocytes) and interorgan cross-talk (gut, liver, and adipose tissue) coordinately maintain energy homeostasis. The physiological synergy between IL-1β and FGF21 in lipolysis might be needed to cope with the concomitant challenge of nutrients.

We thank Dr. Honglin Wang for providing CCR2^−/− mice. We thank Dr. Steven Kliewer for providing FGF21 KO mice. We thank the Flow Cytometry Facility, Animal Facility, and Imaging Facility of the National Center for Protein Science for their assistance.

The authors have no financial conflicts of interest.